Create Word Tree

• normalized lemma or phrase
• original source form when normalization changed it
• meaning
• grammar forms that apply
• provisional or verified CEFR level
• daily CEFR file path
• cumulative CEFR file path

17. Group entries by CEFR level and create one dated file per level at: vocab/<CEFR>/daily/vocab-<CEFR>-YYYY-MM-DD.md
18. Update the cumulative CEFR file for each level at: vocab/<CEFR>/vocab-<CEFR>.md
19. If a dated target file already exists, do not overwrite it unless the user explicitly asks to regenerate it.
20. When updating the cumulative CEFR file, add only new words for that level. Do not duplicate entries that are already present.
21. After creating, regenerating, or skipping output files, append a summary entry to ../CHANGELOG.md by running: ../scripts/append-skill-log.sh "create_word_tree" "<language>" "<summary>"

CEFR Lookup Rules

• Use web search for each entry because CEFR labels are source-dependent and can change across dictionaries.
• Prefer dictionary or language-learning sources that explicitly publish CEFR labels.
• If multiple sources disagree, choose the level supported by the strongest explicit source and briefly note the ambiguity.
• Make it clear when a CEFR level is inferred from partial evidence rather than directly labeled by the source.

Output Structure

Each generated dated CEFR file must include:

• A title with the CEFR level and date
• Source Note
• Level Summary
• One section per part of speech present in that file

Each cumulative CEFR file must include:

• A title with the CEFR level
• Level Summary
• Source Files
• One section per part of speech present in that level

Each vocab entry should be rendered as a compact, readable block that includes:

• the main word or phrase
• article first, if applicable
• part of speech
• the grammar forms that apply
• a short CEFR note with the source or confidence

The rewritten daily note must include:

• the original title with the date
• Raw Capture
• Normalized Word Tree
• one section per part of speech present in the source

Generation Rules

• Cover every usable vocab item from the source note. Do not silently drop entries.
• Cover every usable vocab item in both the rewritten daily note and the CEFR-split files.
• Keep the per-day CEFR files under vocab/<CEFR>/daily/.
• Keep the cumulative CEFR files at vocab/<CEFR>/vocab-<CEFR>.md.
• When a new item belongs to a level, update both the per-day file and the cumulative file for that level.
• Keep entries grouped by part of speech inside each CEFR file.
• Keep entries grouped by part of speech inside the rewritten daily note as well.
• Preserve the original source wording when the note contains a phrase, but still classify it as precisely as possible.
• Do not fabricate forms that are not valid for the entry type.
• When the source note is empty or missing, report that clearly and do not create misleading vocab files or a misleading normalized rewrite.
• The changelog summary should include the target date, the source note path, whether the daily note was rewritten, and which CEFR files were created, regenerated, or skipped.